Embryonic Chick Heart Fragments Research Articles

Chick Heart Invasion Assay for Testing the Invasiveness of Cancer Cells and the Activity of Potentially Anti-invasive Compounds.

The goal of the chick heart assay is to offer a relevant organ culture method to study tumor invasion in three dimensions. The assay can distinguish between invasive and non-invasive cells, and enables study of the effects of test compounds on tumor invasion. Cancer cells - either as aggregates or single cells - are confronted with fragments of embryonic chick heart. After organ culture in suspension for a few days or weeks the confronting cultures are fixed and embedded in paraffin for histological analysis. The three-dimensional interaction between the cancer cells and the normal tissue is then reconstructed from serial sections stained with hematoxylin-eosin or after immunohistochemical staining for epitopes in the heart tissue or the confronting cancer cells. The assay is consistent with the recent concept that cancer invasion is the result of molecular interactions between the cancer cells and their neighbouring stromal host elements (myofibroblasts, endothelial cells, extracellular matrix components, etc.). Here, this stromal environment is offered to the cancer cells as a living tissue fragment. Supporting aspects to the relevance of the assay are multiple. Invasion in the assay is in accordance with the criteria of cancer invasion: progressive occupation and replacement in time and space of the host tissue, and invasiveness and non-invasiveness in vivo of the confronting cells generally correlates with the outcome of the assay. Furthermore, the invasion pattern of cells in vivo, as defined by pathologists, is reflected in the histological images in the assay. Quantitative structure-activity relation (QSAR) analysis of the results obtained with numerous potentially anti-invasive organic congener compounds allowed the study of structure-activity relations for flavonoids and chalcones, and known anti-metastatic drugs used in the clinic (e.g., microtubule inhibitors) inhibit invasion in the assay as well. However, the assay does not take into account immunological contributions to cancer invasion.

The role of non-muscle myosin IIA in aggregation and invasion of human MCF-7 breast cancer cells

Human MCF-7/6 breast cancer cells differ from their MCF-7/AZ counterparts by their invasiveness in a number of assays in vitro, such as invasion of MCF-7 spheroids into embryonic chick heart fragments or type I collagen gels. Comparative proteomic analysis of these two variants revealed an identical pattern, except for a 230 kDa protein present in the invasive MCF-7/6 variant, but hardly detectable in the non-invasive MCF-7/AZ one. This protein appeared to be the non-muscle myosin IIA heavy chain (NMIIA), also coined MYH9. Experimental inhibition of NMIIA by reducing either its expression (via stable shRNA transduction) or its function (via the specific ATPase inhibitor blebbistatin) underpinned the decisive role of NMIIA in MCF-7 cell invasion. Inhibition of NMIIA indeed blocked the invasion of MCF-7/6 cells in three-dimensional invasion substrata such as embryonic chick heart fragments and type I collagen gels. Invasiveness of MCF-7/6 cells has been related to poor formation and compaction of aggregates, due to a functionally defective E-cadherin/catenin complex. Both genetic and pharmacological inhibition of NMIIA stimulated MCF-7/6 cell aggregation. Together, these data indicate that NMIIA is a decisive protein for MCF-7 cells to invade, indicating that this molecule is a candidate for targeted anti-invasive treatment.

Synthesis, characterization and in vitro anti-invasive activity screening of polyphenolic and heterocyclic compounds

Invasion is the hallmark of malignant tumors, and is responsible for the bad prognosis of the untreated cancer patients. The search for anti-invasive treatments led us to screen compounds of different classes for their effect in an assay for invasion. Thirty-nine new compounds synthesized in the present study along with 56 already reported compounds belonging mainly to the classes of lactones, pyrazoles, isoxazoles, coumarins, desoxybenzoins, aromatic ketones, chalcones, chromans, isoflavanones have been tested against organotypic confronting cultures of invasive human MCF-7/6 mammary carcinoma cells with embryonic chick heart fragments in vitro. Three of them (a pyrazole derivative, an isoxazolylcoumarin and a prenylated desoxybenzoin) inhibited invasion at concentrations as low as 1 μM; instead of occupying and replacing the heart tissue within 8 days, the MCF-7/6 cells grew around the heart fragments and left it intact, when treated with these compounds. At the anti-invasive concentration of 1 μM, the three compounds did not affect the growth of the MCF-7/6 cells, as shown in the sulforhodamine B assay. Aggregate formation on agar was not stimulated by any of the three anti-invasive compounds, making an effect on the E-cadherin/catenin complex improbable. This is an invasion suppressor that can be activated in MCF-7/6 cells by a number of other molecules. Our data indicate that some polyphenolic and heterocyclic compounds are anti-invasive without being cytotoxic for the cancer cells.

Synthetic and biological activity evaluation studies on novel 1,3-diarylpropenones

Fourteen novel C-prenylated and O-allylated 1,3-diarylpropenones (chalcones) were synthesized by Claisen–Schmidt condensation reaction of C-prenylated/O-allylated acetophenones with appropriate aldehydes; twelve of these model chalcones were screened in an assay based on the confrontation of invasive human MCF-7/6 mammary carcinoma cells with fragments of normal embryonic chick heart in vitro. Out of the twelve chalcones tested, three were found to exhibit potent anti-invasive activity. Some of these chalcones and their precursor acetophenones were also tested for inhibition of initiation of lipid peroxidation in rat liver microsomes; a prenylated acetophenone carrying two methoxy groups and two free phenolic hydroxy functions was found to be a potential antioxidant.

Models for studying cellular invasion of basement membranes.

Invasion of extracellular matrices is crucial to a number of physiological and pathophysiological states, including tumor cell metastasis, arthritis, embryo implantation, wound healing, and early development. To isolate invasion from the additional complexities of these scenarios a number of in vitro invasion assays have been developed over the years. Early studies employed intact tissues, like denuded amniotic membrane (1) or embryonic chick heart fragments (2), however recently, purified matrix components or complex matrix extracts have been used to provide more uniform and often more rapid analyses (for examples, see the following integrin studies). Of course, the more holistic view of invasion offered in the earlier assays is valuable and cannot be fully reproduced in these more rapid assays, but advantages of reproducibility among replicates, ease of preparation and analysis, and overall high throughput favor the newer assays. In this chapter, we will focus on providing detailed protocols for Matrigel-based assays (Matrigel=reconstituted basement membrane; reviewed in ref. (3)). Matrigel is an extract from the transplantable Engelbreth-Holm-Swarm murine sarcoma that deposits a multilammelar basement membrane. Matrigel is available commercially (Becton Dickinson, Bedford, MA), and can be manipulated as a liquid at 4°C into a variety of different formats. Alternatively, cell culture inserts precoated with Matrigel can be purchased for even greater simplicity.

Protein Kinase C-α Activity Inversely Modulates Invasion and Growth of Intestinal Cells

The phorbol ester phorbol 12-myristate 13-acetate induces remarkable phenotypic changes in intestinal HT-29 M6 cells; these changes consist of loss of homotypic adhesion and inactivation of E-cadherin. In parallel, cell growth is retarded. We have transfected HT-29 M6 cells with an activated form of the conventional protein kinase Calpha (cPK-Calpha). Expression of this isoform induced the acquisition of a scattered phenotype, similar to that adopted by cells after addition of phorbol 12-myristate 13-acetate, with very low cell-to-cell aggregation and undetectable levels of functional E-cadherin. These cell clones were highly motile and rapidly invaded embryonic chick heart fragments. Furthermore, cells expressing activated-cPK-Calpha showed decreased proliferation in comparison to control clones. We have also studied how these two apparently antagonistic changes affect the tumorigenic ability of HT-29 M6 cells. When the different cell clones were xenografted into athymic mice, the effect on cell growth seemed to predominate. Expression of activated-cPK-Calpha significantly reduced the size of the tumors; the cells with the highest level of expression did not even form subcutaneous tumors. Besides their smaller size, the morphology of these tumors was clearly different from those originated by HT-29 M6 cells, and they could be defined as infiltrative on anatomo-pathological basis. These results indicate that cPK-Calpha controls both cell-to-cell adhesion and proliferation of intestinal cells.

Anti-invasive activity of alkaloids and polyphenolics in vitro

Invasiveness, the ability of certain tumour cells to migrate beyond their natural tissue boundaries, often leads to metastasis, and usually determines the fatal outcome of cancer. The need for anti-invasive agents has led us to search for possibly active compounds among alkaloids and polyphenolics. One hundred compounds were screened in an assay based on the confrontation of invasive human MCF-7/6 mammary carcinoma cells with fragments of normal embryonic chick heart in vitro. Anti-invasive activity was frequently found among chalcones having a prenyl group. Six compounds were found to inhibit invasion when added to the culture medium at concentrations as low as 1 μM. For at least three of them the anti-invasive effect could be associated with a cytotoxic effect on the MCF-7/6 cells, but not on the heart tissue. This selective cytotoxicity was substantiated by different methods, such as histology and growth assays (volume measurements, cell counts, MTT and sulforhodamine B assays). The anti-invasive effects of the compounds could neither be ascribed to induction of apoptosis nor to the promotion of cell-cell adhesion. Our data indicate that among the alkaloids and polyphenolics a number of molecules can inhibit growth and invasion of human mammary cancer cells via selective cytotoxicity.

In vivo and in vitro invasion in relation to phenotypic characteristics of human colorectal carcinoma cells

In this study we investigated the tumorigenicity, growth pattern and spontaneous metastatic ability of a series of nine human colorectal carcinoma cell lines after subcutaneous and intracaecal xenografting in nude mice. CaCo2 cells were found to be poorly tumorigenic to non-tumorigenic in either site; the other cell lines were tumorigenic in both sites. SW1116, SW480 and SW620 did not show local invasive in the NCI-H716 and LS174T cells were both invasive in the caecum, but only NCI-H716 was invasive in the subcutis. HT29 and 5583 (S and E) cells were invasive in the caecum and from that site metastatic to the lungs and/or the liver. HT29 and 5583S cells were both invasive in the subcutis, but 5583E cells were not. Of each category of in vivo behaviour in the caecum, one cell line was further investigated with regard to invasion in vitro (into embryonic chick heart fragments), E-cadherin expression in vivo and in vitro and in vitro production of u-PA and t-PA. These parameters were chosen in view of their purported role in extracellular matrix degradation and intercellular adhesion, which are all involved in the invasive and metastatic cascade. Invasion in vitro was not predictive for invasion or metastasis in vivo. In the cell line which showed invasion in embryonic chick heart tissue, heterogeneous E-cadherin expression was observed in vitro together with a relatively high production of u-PA. The non-invasive cell lines showed in vitro homogeneous expression of E-cadherin with a relatively low production of u-PA. In vivo expression of E-cadherin was either absent or heterogeneous. We conclude that: (1) colorectal carcinoma xenografts show site-specific modification of in vivo invasive and metastatic behaviour; (2) invasion in vitro does not correlate with invasion and metastasis in vivo; (3) in vitro non-invasion might be associated with homogeneous E-cadherin expression and low production of u-PA; (4) E-cadherin expression in vitro differs from E-cadherin expression in vivo. The results support the notion that the microenvironment in which cancer cells grow is one of the factors involved in the regulation of invasive and metastatic behaviour.

Anti-Invasive Activity of 3,7-Dimethoxyflavone in Vitro

Invasion of MCF-7/6 human mammary carcinoma cells into embryonic chick heart fragments was studied in organ culture during 8 days. The effect of 31 polyphenolic compounds, belonging to the flavonoids, chalcones, or coumarins, was tested in this assay for invasion. The anti-invasive activity of 3,7-dimethoxyflavone was found at concentrations ranging from 1 to 100 μM. At these anti-invasive concentrations, no cytotoxic effects could be detected: the anti-invasive effect was reversible upon omission of the molecule from the medium, and treatment of MCF-7/6 cells or heart fragments did not affect subsequent outgrowth from explants on tissue culture plastic. The molecule did not inhibit growth of MCF-7/6 cell aggregates nor of heart fragments kept in suspension culture. The action mechanism of 3,7-dimethoxyflavone is the subject of our ongoing research.

Adhesion and invasion of three human choriocarcinoma cell lines into human endometrium in a three-dimensional organ culture system

A novel in vitro model was developed to study attachment and invasion of choriocarcinoma cell spheroids using pre-cultured secretory phase human endometrium as a host tissue. During pre-culturing in shaker culture human endometrium had regenerated a complete epithelial covering and had shed cells damaged during explantation. Spheroids of three human choriocarcinoma cell lines (BeWo, Jeg-3, JAr) which displayed linear growth in culture and produced placental hormones were used in this study as models for trophoblast behaviour. Morphological differences were noted in the spheroids from the three choriocarcinoma cell lines; BeWo and Jeg-3 spheroids exposed flattened and more differentiated cells on their surfaces while superficial cells in JAr spheroids maintained their cytotrophoblast-like morphology. Spheroids from all three cell lines were proven to be invasive in a general invasion assay using embryonic chick heart fragments, with JAr spheroids being the most aggressive. When spheroids were confronted with pre-cultured re-epithelialized endometrial fragments, however, Jeg-3 spheroids showed the highest incidence of attachment (52%) and the greatest amount of invasion into the underlying stroma. BeWo spheroids also attached (37%) and penetrated the epithelium, but did not invade into the stroma. JAr spheroids showed a minor degree of attachment (12%) and little or no invasion into the stroma. These results show that the three choriocarcinoma cell lines, although all invasive in a general invasion assay, differ in adhesion to uterine epithelium and invasion into endometrial stroma. This model offers opportunities for studying mechanisms of trophoblast adhesion and invasion, using human endometrium as the natural host tissue.

Insulin-like growth factor I activates the invasion suppressor function of E-cadherin in MCF-7 human mammary carcinoma cells in vitro.

The calcium-dependent cell-cell adhesion molecule E-cadherin has been shown to counteract invasion of epithelial neoplastic cells. Using three monoclonal antibodies, we have demonstrated the presence of E-cadherin at the surface of human MCF-7/6 mammary carcinoma cells by indirect immunofluorescence coupled to flow cytometry and by immunocytochemistry. Nevertheless, MCF-7/6 cells failed to aggregate in a medium containing 1.25 mM CaCl2, and they were invasive after confrontation with embryonic chick heart fragments in organ culture. Treatment of MCF-7/6 cells with 0.5 microgram ml-1 insulin-like growth factor I (IGF-I) led to homotypic aggregation within 5 to 10 min and inhibited invasion in vitro during at least 8 days. The effect of IGF-I on cellular aggregation was insensitive to cycloheximide. However, monoclonal antibodies that interfered with the function of either the IGF-I receptor (alpha IR3) or E-cadherin (HECD-1, MB2) blocked the effect of IGF-I on aggregation. The effects of IGF-I on aggregation and on invasion could be mimicked by 1 microgram ml-1 insulin, but not by 0.5 microgram ml-1 IGF-II. The insulin effects were presumably not mediated by the IGF-I receptor, since they could not be blocked by an antibody against this receptor (alpha IR3). Our results indicate that IGF-I activates the invasion suppressor role of E-cadherin in MCF-7/6 cells.

Modulation of invasive potential in different clonal subpopulations of a rat rhabdomyosarcoma cell line (BA-HAN-1) by differentiation induction

Three clonal subpopulations (A, B, C) isolated from the same rhabdomyosarcoma of the rat and differing in their degree of spontaneous differentiation were tested for their invasive potential before and after differentiation induction with retinoic acid (RA), N-monomethylformamide (NMF) and sodium butyrate (NaBut). Invasive potential was analysed in an in vitro assay using embryonic chick heart fragments in organotypic culture. In standard culture medium, all three subpopulations were shown to be invasive, progressively replacing the chick heart fragments within 7-11 days after confrontation. After exposure to RA, NMF or NaBut, marked differences in the invasive potential of these subpopulations were, however, observed. Subpopulation C exhibited a pronounced decline in invasive potential, as evidenced by a significant decrease (P = 0.005) in the proportion of chick heart fragments with advanced stages of invasion. This response, however, was confined to the differentiation-inducing agents RA and NaBut, which had also produced a marked increase in morphological and/or biochemical differentiation (P = 0.0001). In contrast, NMF, which had only minor effects on differentiation, failed to affect the invasive potential of subpopulation C. In subpopulation B, a transient inhibition of single cell invasion became evident after exposure to RA, whereas NMF and NaBut failed to affect the invasive potential in spite of minor effects on differentiation. In the least differentiated subpopulation A, which was shown to be refractory to the differentiation-inducing effects of RA, NMF and NaBut, there was also no observation of any reduction of invasive potential. The results of our study demonstrate that differentiation-inducing agents can significantly reduce the invasive potential of malignant tumors, although marked differences of response are to be expected between the different subpopulations of a tumor.

INVASIVENESS INVITRO VERSUS CLINICAL EVOLUTION OF PRIMARY AND SECONDARY BRAIN-TUMORS

Cells of 56 primary cell cultures derived from primary and secondary brain tumours were confronted in vitro with host tissue, embryonic chick heart fragments (PHF). The evolution of the confrontation was followed for one week. Histological analysis of the confrontations demonstrated two different patterns. (i) Survival of host tissue and tumour derived cells without invasion of the last ones into the host. (ii) Invasion of tumour derived cells into the host tissue followed by replacement of the chick heart tissue. Invasion into the host tissue was observed only in the confrontations with primary cell cultures originating from malignant and metastatic tumours. The invasive characters in vitro of primary cell cultures correlated with the malignancy of the tumour in vivo.

Invasiveness in in vitro and clinical evaluation of meningiomas

Fragments of freshly isolated human meningiomas were cultured in vitro to form cell monolayers. These monolayers were confronted with embryonic chick heart fragments in vitro for 1, 2, 4, and 7 days. Microscopically, three different histological patterns were observed. Type I included necrotized meningial cells; type II presented surviving meningial cells; type III included meningial cells that had invaded the host tissue. The clinical analysis included the histopathological diagnosis, the macroscopic situation at surgical intervention, and the follow-up with or without recurrence. Correlation between these clinical parameters and the in vitro results demonstrated that type III confrontations correlated with macroscopic infiltration in the brain parenchyma and tumor recurrence. Invasiveness in vitro was seen in two anaplastic and two transitional meningiomas.

Retinoic acid modulates both invasion and plasma membrane ruffling of MCF-7 human mammary carcinoma cells in vitro.

The invasiveness of MCF-7 human mammary carcinoma cells was tested in vitro via confronting cultures with embryonic chick heart fragments. Invasive (e.g. MCF-7/6) and non-invasive (e.g. MCF-7/AZ) variants were detected. Automated image analysis of time-lapse video-microscopy recordings showed that the plasma membrane ruffling activity of the invasive MCF-7/6 variant was higher than the ruffling activity of the non-invasive MCF-7/AZ variant. Addition of all-trans-retinoic acid to the culture medium (10(-6) M) inhibited both invasion and ruffling of MCF-7/6 cells, while MCF-7/AZ cells became invasive and acquired an increased ruffling by the same type of treatment. A similar opposite effect on MCF-7 cells was not found after treatment with other ligands of the nuclear steroid/thyroid receptor superfamily. Triiodo-l-thyronine (up to 10(-5) M) and beta-oestradiol (up to 10(-6) M) did not alter the invasiveness of the cells, while dexamethasone (10(-6) M) and the pure anti-oestrogen ICI 164,384 inhibited both invasion and ruffling. Our data show that retinoic acid can modulate invasiveness in opposite directions.

Attachment and invasion of high- and low-metastatic clones of RCT sarcoma in a three-dimensional culture system

High- and low-metastatic clones established from poorly differentiated murine sarcoma (RCT sarcoma) spontaneously developed in C3H/He mice were used in this study. Attachment and invasion, which are factors likely related to metastatic ability, of these clones were investigated in a three-dimensional culture system using embryonic chick heart fragments and tumor cell aggregates. The effects of Nocodazole, which interferes with the function of microtubules, and cytochalasin B, which affects the actin assembly in cytoplasmic extensions, on these factors were also examined. Metastatic ability was correlated with attachment and invasiveness of RCT sarcoma. Invasiveness of tumor cells was suppressed by the addition of Nocodazole to the culture medium. Since the disappearance of cell polarity, resulting from the inhibition of tubulin assembly, was found to coincide with the suppression of invasion, directional migration is considered to be involved in the invasion of tumor cells. Cytochalasin B caused a marked decrease in the ratio of attachment of tumor aggregates to embryonic chick heart fragments. The increased ability of invasion and attachment of RCT(+) cells compared with RCT(-) cells seems to contribute to their propensity to metastasize.

Effect of catechins and citrus flavonoids on invasionin vitro

Catechins, a group of flavonoid molecules, inhibit invasion of mouse MO4 cells into embryonic chick heart fragments in vitro. The anti-invasive effects can be ranked as follows: (+)-catechin greater than (-)-epicatechin greater than 3-O-methyl-(+)-catechin greater than 3-O-palmitoyl-(+)-catechin. Most of the catechins are unstable in cell culture media, and their spontaneous rearrangement products tend to bind to extracellular matrix (ECM). Due to these interactions proteases such as tissue-type plasminogen activator (t-PA) are linked to the ECM glycoprotein laminin. This leads to a partial inactivation of the enzyme. Within the group of catechins we found a positive correlation between anti-invasive activity and linking of t-PA to laminin. Citrus flavonoids are also anti-invasive in vitro (tangeretin greater than nobiletin greater than hesperidin = naringin). However, these stable molecules show poor affinity for ECM, and do not link enzymes to laminin. These data suggest that catechins and citrus flavonoids inhibit invasion in vitro by different mechanisms.

Newly established human retinoblastoma cell lines exhibit an “immortalized” but not an invasive phenotype in vitro

Retinoblastoma (RB), an intraocular childhood tumor occurring in a hereditary (mostly bilateral) or non-hereditary (unilateral) form, is associated with the inactivation of both alleles of a putative tumor suppressor gene (RB-I) located on chromosome 13q14. Both the process of RB development and the biological characteristics of RB cells are as yet poorly understood. We have established 7 new RBL lines (RBL13, RBL14, RBL18 and RBL30, derived from unilateral RB; and RBL7, RBL15 and RBL20, derived from bilateral RB). Southern blot analyses of restriction fragment length polymorphisms in DNA samples from 6 cell lines revealed loss of constitutional heterozygosity at one or several polymorphic loci on chromosome 13 in 4 cases. Gross deletions involving the RB-I locus and amplification of the N-myc gene were not detected in any of the RBL lines. The phenotypic properties of the RBL lines were analyzed in comparison with cells from the original RB tumors, with 4 RB lines established by others (RB383, RB355, RB247C3 and Y79) and with the adenovirus-EIA-transformed human retinoblast line HER-Xhol-CC2. It was found that RB tumors consist of phenotypically heterogeneous cell subpopulations with varying nutrient requirements and differentiation potential in vitro. All cell lines showed the typical characteristics of established ("immortalized") cells. In some cases, cells from original RB tumors or cell lines were able to form colonies when cell aggregates of 2-10 cells were suspended in semi-solid agar medium; however, anchorage-independent colonies never developed from single cells. Cell lines RBL13, RBL18, RB247C3, RB355, RB383 and Y79 were tested for invasion into embryonic chick heart fragments in vitro and found to be non-invasive. None of the RBL or RB lines were tumorigenic in nu/nu (T-) mice. Y79 cells (propagated in culture for many years) exhibited properties distinctly different from those of the other cell lines, and thus cannot be considered phenotypically representative of RB cells.

Invasion of human brain tumors in vitro: relationship to clinical evolution

Embryonic chick heart fragments were confronted in vitro with cells from 26 freshly resected human brain tumors. The tumor-derived cells were scored according to their survival and invasiveness. Four different responses were observed: disintegration of the tumor-derived cells (Type I); survival of cells without progressive engulfing or invasion of the heart fragment (Type II); initial encircling of the heart fragment followed by invasion (Type III); and massive invasion on initial contact (Type IV). Pattern Type III or IV was seen in 11 of 14 preparations derived from malignant tumors, and pattern Type I or II was seen in seven of 10 preparations derived from benign tumors. These results suggested that "classical malignancy," based upon the histological classification and the degree of spread in vivo, correlated with invasiveness in vitro.

Quantitative Evaluation of Melanoma Cell Invasion in Three-Dimensional Confrontation Cultures In Vitro Using Automated Image Analysis

Tumor invasion is a crucial feature of tumor growth in vivo. Confrontation cultures of multicellular melanoma spheroids and embryonic chick heart fragments provide a model for invasive growth in vitro. We have developed an image analysis method, which facilitates the objective measurement of tumor cell invasion in this model. Cryostat sections of confrontation cultures were immunohistochemically stained with an antiserum directed against the stromal component for automated recognition of the stroma tissue. The slides were automatically processed by a grey level based computerized image analysis system. On Spearman's rank correlation test, 25 out of 39 parameters correlated with the reference value of invasion, which was derived from the subjective evaluation of five independent observers. Two parameters combining the stroma margin and the total amount of stroma tissue completely reproduced the judgement of the morphologists in our test set. The quantitative evaluation of tumor invasion in vitro by automated image analysis may be helpful in pharmacologic and pathogenetic studies of tumor growth.